Ksepca et al, 2006

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY

CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024

Number 3508, 16 pp., 12 figures, 3 tables March 16, 2006

Erketu ellisoni, a Long-Necked Sauropod from BorGuve (Dornogov Aimag, Mongolia)

DANIEL T. KSEPKA1 AND MARK A. NORELL2

ABSTRACT

The first specimen of the new sauropod Erketu ellisoni, from the Lower Cretaceous of the easternGobi of Dornogov, Mongolia, is described here. The specimen comprises a well-preservedarticulated anterior cervical series, an articulated lower hindlimb, and a sternal plate. Thissauropod displays a unique combination of features including low, bifid neural spines, elongatecervical centra, and crescent-shaped sternal plates. Computed tomography imaging reveals thevertebrae were extensively invaded with pneumatic camellae. The holotype individual of Erketuwas of modest mass relative to other neosauropods, but had an extremely elongate neck.Phylogenetic analysis indicates Erketu is a member of the Somphospondyli and may belong toa more exclusive clade therein.

INTRODUCTION

The 2002 American Museum of NaturalHistory–Mongolian Academy of Sciences fieldexpedition discovered a new locality, BorGuve, in Dornogov Aimag, Mongolia (fig. 1).Several sauropod cervical vertebrae werefound eroding out of the surface at thislocality, and excavation revealed the rest ofthe anterior cervical series preserved in artic-ulation directly in front of these elements.Additional postcranial elements were uncov-ered as the cervicals were being collected.Despite exhaustive efforts, no skull was foundin the vicinity of the atlas.

The first sauropod remains described fromMongolia were two teeth collected by the

American Museum of Natural HistoryCentral Asiatic Expeditions at the LowerCretaceous Ooshiin Nuur locality (Osborn,1924). These teeth were assigned toAsiatosaurus mongoliensis, a name now con-sidered a nomen dubium (McIntosh, 1990).The Central Asiatic Expeditions also collecteda basioccipital, three cervical vertebrae, andseveral teeth of the poorly known sauropodMonoglosaurus haplodon from the LowerCretaceous Hu Khung Ulan locality(Gilmore, 1933). Several important sauropodspecimens have since been recovered fromUpper Cretaceous deposits, including thenearly complete postcranial skeleton ofOpisthocoelicaudia (Borsuk-Bialynicka, 1977)and skull of Nemegtosaurus, collected by the

1Division of Paleontology, American Museum of Natural History ([email protected]).2Division of Paleontology, American Museum of Natural History ([email protected]).

Copyright E American Museum of Natural History 2006 ISSN 0003-0082

Polish–Mongolian expeditions (Nowinski,1971), and the skull of Quesitosaurus, col-lected by the Soviet–Mongolian expeditions(Kurzanov and Bannikov, 1983). With theexception of these discoveries, sauropod ma-terial from Mongolia has been scarce andfragmentary. The discovery of Erketu providesan important addition to Mongolian sauropoddiversity.

The deposits at Bor Guve appear torepresent a floodplain environment. Inter-bedded grey siltsones and channel sandstonesdominate. Lag deposits have yielded turtleremains and dinosaur teeth, and sauropod andtheropod remains have been recovered fromthe siltsones. Among the theropod materialuncovered at this site are elements froma maniraptoran slightly larger thanDeinonychus and an Allosaurus-sized predato-ry theropod. Fossil fruits (fig. 2) froma probable angiosperm plant of unknownaffinity are locally abundant at the locality.These fruits bear a superficial resemblance toextant Abelmoschus esculentus (okra). Thefruits are not, to our knowledge, assignableto any known taxon, and are currently

being described (Koppelhus, in prep.).Indistinguishable fossil fruits have been col-lected from the stratigraphically lower KharaKhuutal beds, which crop out to the east.

In describing the morphology of the verte-brae, we use the terminology and abbrevia-tions for vertebral lamina proposed by Wilson(1999) and the terminology for pneumaticfeatures proposed by Wedel et al. (2000).Computed tomography (CT) scans were per-formed at the Stony Brook UniversityHospital Radiology Department, using a GELight Speed 16 scanner.

ABBREVIATIONS

INSTITUTIONAL: AMNH, American Muse-um of Natural History, New York, NewYork, USA; CM, Carnagie Museum ofNatural History, Pittsburgh, Pennsylvania,USA; CCG, Chengdu College of Geology,Sichuan, China; HM, Humbolt Museum,Berlin, Germany; IGM, Geological Instituteof the Mongolian Academy of Sciences, UlaanBaatar, Mongolia; OMNH, OklahomaMuseum of Natural History, Norman,

Fig. 1. Map of Mongolia showing the location of Bor Guve.

2 AMERICAN MUSEUM NOVITATES NO. 3508

Oklahoma, USA; PMU, PaleontologicalMuseum, Uppsala, Sweden; ZDM, ZigongDinosaur Museum, Zigong, China.

ANATOMICAL: ap, accessory anterior pro-cess of prezygodiapophyseal lamina; ax, axis;cml, camella; cr, cervical rib; ic, intercentrumof atlas; mtp, metapophysis; np, neurapophysisof atlas; pcdl, posterior centrodiapophyseallamina; pcpl, posterior centroparapophyseallamina; pdol, postzygodiapophyseal lamina;pfm, pneumatic foramen; pfs, pneumatic fossa;prdl, prezygodiapophyseal lamina; prz, pre-zygapophysis; tb, tubercle; td, torus dorsalis,tp, transverse process.

SYSTEMATIC PALEONTOLOGY

SAUROPODA MARSH, 1878

NEOSAUROPODA BONAPARTE, 1986

MACRONARIA WILSON AND SERENO, 1998

TITANOSAURIFORMES SALGADO, CORIA, AND

CALVO, 1997

SOMPHOSPONDYLI WILSON AND SERENO, 1998

Erketu ellisoni new taxonHOLOTYPE: IGM 100/1803: articulated cer-

vical series including complete first throughfifth cervical vertebrae, partial sixth cervicalvertebra, right sternal plate, articulated righttibia, fibula, astragalus, and calcaneum.

ETYMOLOGY: Erketu: In Mongolian sha-manistic tradition, there are 99 Tengri (de-ities). Erketu Tengri is the Mighty Tengri,a creator-god who called Yesugei, the fatherof Chingis Khan, into being. ellisoni: In honor

of Mick Ellison, for his contributions toongoing AMNH dinosaur research.

DIAGNOSIS: Referable to Titanosauriformesbased on elongate cervical vertebrae withcamellae and referable within Titanosauri-formes to Somphospondyli based on reducedneural arch lamination. Differentiated fromall other Titanosauriformes in which cervicalvertebrae are known by combination ofextremely elongated (EI indices of anteriorcervicals exceeding 5.0) cervical centra andbifurcate anterior cervical neural spines.

TYPE LOCALITY AND HORIZON: Bor Guve:late Early Cretaceous. Stratigraphically, thebeds at this locality lie below the TsaaganTsonch beds (a unit that contains Iguanodonorientalis) and above the Khara Khuutul beds,both believed to be of late Early Cretaceousage (Shuvalov, 2000). The lack of materialssuitable for radiometric dating leaves the exactage of this locality uncertain.

DESCRIPTION

AXIAL SKELETON

The atlas is complete and articulated withthe axis (fig. 3). The intercentrum is cresecen-tic in anterior view, though deformation hascompressed the dorsal apices inward, causingit to appear more circular. In lateral view, theintercentrum is subrectangular. A roughenedarticular surface is present near the poster-oventral corner of the lateral face. The wing-like neurapophyses embrace the neural spineof the axis. The fusion of the neurapophyses tothe intercentrum and closure of all otherneurocentral sutures indicates IGM 100/1803represents an adult individual.

Fig. 2. Fossil fruits from Bor Guve. Scale bar 5 1 cm.

2006 KSEPKA AND NORELL: SAUROPOD FROM MONGOLIA 3

The axis is opisthocoelous with well-de-veloped pneumatic fossae. The anterior half ofthe axial centrum develops a strong ventralkeel. The cotyle is subcircular, with nearlyequal height and width. The short transverseprocess has a ventrally directed distalflange. The anteriorly displaced parapophysesbear single-headed ribs that extend signifi-cantly beyond the posterior border of theaxial centrum, though the total length isuncertain due to breakage. The axial neuralspine barely projects dorsal to the level of thepostzygapophyses. The low axial neuralspine of Erketu is similar to that of theunnamed titanosaur from Peiropolis, Brazil,referred to as the ‘‘Series A’’ taxon (Powell,2003) and the eusauropod Mamenchisaurussinocandorum (Russell and Zheng, 1993; no-tably, the axial neural projects well dorsalto the postzygapophyses in the supposedlycogeneric Mamenchisaurus houchuanensis;

Young and Zhou, 1972). Most titanosauri-forms, including Brachiosaurus (Janesch, 1950:fig. 14), Euhelopus (Wiman, 1929: pl. 3, fig. 3),and Saltasaurus (Powell, 1992) have moredorsally projected axial neural spines.

The third (fig. 4) and succeeding cervicalvertebrae have opisthocoelous centra withwell-developed pneumatic fossae and variablydeveloped laminae. The most striking featureof the cervical vertebrae is their elongation.The elongation index (EI, sensu Wedel et al.,2000: centrum length/cotyle height) exceedsthose of all other sauropods for whichcomparable data are available (table 1). Itshould be noted that because the cotyle istaller than wide in Erketu, calculating EI ascentrum length/cotyle width (sensu Upchurch,1998) would increase the values for this taxonboth absolutely and relative to most othersauropods. Sauroposeidon proteles (Wedelet al., 2000) and Omeisaurus junghsiensis

Fig. 3. Atlas and axis of IGM 100/1803 in (a) right lateral, (b) left lateral, (c) dorsal, (d) ventral, (e)anterior, and (f) posterior view. Note that the cervical rib has been detached in (c)–(f). For abbreviations, seetext. Scale bar for (a), (b), (c), and (d) 5 5 cm.


(Young, 1939) also possess tremendouslyelongate vertebrae. However, only mid-cervi-cals are preserved for S. proteles, and in O.junghsiensis the preserved cervicals have suf-fered distortion of the cotyles and are of

uncertain position, precluding direct compar-ison with Erketu.

As in Euhelopus and the primitive Asianeusauropods Shunosaurus, Omeisaurus, andMamenchisaurus, the postaxial centra are

Fig. 4. Cervical 3 of IGM 100/1803 in (a) left lateral, (b) right lateral, (c) dorsal, (d) anterior, and (e)posterior view. For abbreviations, see text. Scale bar for (a), (b), and (c) 5 5 cm.


significantly higher than wide at the cotyle(centrum six is wider than high due todorsoventral deformation). The opposite con-dition is developed in the basal titanosaurPhuwiangosaurus, which has transversely ex-panded cervical vertebral centra (Martin et al.,1994). The lateral surface of each cervicalcentra in Erketu is marked by a large pneu-matic fossa that reduces the centum to a thinmedian septum. The posterior centrodiapo-physeal and posterior centroparapophyseallamina are strongly developed along the areaof the pneumatic fossa, but are weak near theposterior border of the centrum. A weaklydeveloped lamina divides the large lateralpneumatic fossa from a smaller anterior

pneumatic fossa in the anterior cervicalvertebrae. However, this lamina is betterdeveloped in the sixth cervical vertebra, in-dicative of an expected increase in complexityof the fossae moving posterior in the vertebralcolumn. The anterior pneumatic fossa extendsventrally onto the dorsal surface of theparapophyses on the third and fourth cervi-cals, where the parapophyses are preserved.Additional excavations are present on thecentra and neural spines. The distribution ofthese minor excavations, of pneumatic foram-ina in the pneumatic fossa, and of similarfeatures elsewhere on the vertebrae is highlyvariable and may even differ between the twosides of an individual vertebra. The posterior

TABLE 1

Elongation Indices and Centrum Lengths from Various SauropodsValues for Erketu are from this study, values for Mamenchisaurus are from Young and Zhou (1972), valuesfor Omeisaurus tianfuensis are from He et al. (1988), and all other values are from Wedel et al. (2000). ForErketu ellisoni, C5 is dorsoventrally compressed, causing an artificially high EI value of 8.5. If cotyle width isused intead of height, a value of 5.8 is calculated; the true value would be slightly greater.

Vertebra

C2 C3 C4 C5 C6 C7 C8

Centrum length (mm)

Apatosaurus lousisae (CM 3018) 190 280 370 — 440 — 485

Brachiosaurus brancai (HM SII) — 420 663 810 900 930 973

Camarasaurus supremus (AMNH

5761)

235 265 310 395 — 350 605

Diplodocus carnegii (CM 84) 163 243 289 372 442 485 512

Erketu ellisoni (IGM 100/1803) 160 268 387 489 — — —

Euhelopus zdanskyi (PMU.R233) 94 130 222 234 238 260 262

Mamenchisaurus hochuanensis

(CCG V 20401)

160 215 320 415 480 580 590

Omeisaurus tianfuensis (ZDM

T5701)

170 241 368 495 595 670 673

Sauroposeidon proteles (OMNH

53062)

— — — — 1220 1230 1250

Elongation index

Apatosaurus lousisae (CM 3018) 2.2 3.0 3.3 — 3.2 — 2.9

Brachiosaurus brancai (HM SII) — 3.5 4.3 5.4 5.0 4.4 4.0

Camarasaurus supremus (AMNH

5761)

1.9 2.5 2.3 2.3 — 3.1 3.5

Diplodocus carnegii (CM 84) 3.1 4.1 3.3 4.7 4.7 4.9 4.3

Erketu ellisoni (IGM 100/1803) 3.4 4.5 5.5 X — — —

Euhelopus zdanskyi (PMU.R233) 2.5 2.7 5.4 3.6 3.2 3.2 2.8

Mamenchisaurus hochuanensis

(CCG V 20401)

2.0 2.5 2.6 2.8 2.9 2.9 2.7

Omeisaurus tianfuensis (ZDM

T5701)

2.7 3.0 3.8 4.9 3.6 X X

Sauroposeidon proteles (OMNH

53062)

— — — — 6.1 5.6 4.6


centroparapophyseal laminae make the ven-tral surface of the anterior half of the centrumconcave. Posteriorly, the ventral surface isslightly convex.

The neural arches of cervical vertebrae 3–5are characterized by low, gently curvingprofiles. The prezygapophyses extend anteriorof the condyle as in most sauropods. Theprezygodiapophyseal lamina is unremarkablein cervical three. In the succeeding cervicals,however, it is hypertrophied and extends faranterior of the articular facet as a thin plate,oriented perpendicular to the articular plane(fig. 5). The primitive Saharan eusauropodJobaria also possesses a similar morphology(Sereno et al., 1999: ‘‘accessory anterior pro-cess’’). The postzygapophyses extend onlyslightly beyond the posterior border of thecentrum. The postzygodiapophyseal lamina ispoorly developed and fades from a sharplamina to a low, rounded ridge a shortdistance posterior to the diapophysis. Thepronounced torus dosalis extends posterior tothe postzygapophyseal articular facet in cervi-cals four and five, though not in cervical three.The intrapostzygapophyseal lamina connectto the dorsal surface of the neural canal viaa vertical strut. The neural arch laminae arereduced, but not to the degree seen inMalawisaurus. In lateral view, the cervicalneural arches of Erketu resemble those of the‘‘Series A’’ titanosaur and Malawisaurus moreclosely than the higher and more steeplyangled neural arches of Euhelopus (Wiman,1929: pl. 3, fig. 3) and Brachiosaurus (Janesch,1950: figs. 17, 20, 23).

The neural spines of the fourth andsucceeding cervicals are bifurcated. The meta-pophyses are very low and are thickened andrugose at their highest point. A small tubercleprojects from each metapophysis at theanterior edge of the inflated portion (fig. 6).There is no median tubercle between thesplit neural spines, though this does not ruleout the presence of this feature fartherposterior in the cervical series. As mentionedabove, pneumatic features of the neural spineare highly variable. Cervical four possessesa large, well-defined ovoid excavation that issubdivided by a bony strut on the leftmetapophysis, but no counterpart feature onthe right.

The delicately tapering, double-headed ribsare completely fused to the diapophyses andparapophyses. An anterior projection of the ribextends to the anterior limit of the centrum. Themain body of the most complete rib extendsposteriorly to overlap at least one succeedingcentrum, but it is incomplete at the posteriorend and may have been significantly longer.

Computed tomography imaging reveals de-tails of the internal structure of the fourthcervical vertebra (fig. 4). The centrum isreduced to a median plate by the pneumaticfossa at its midpoint. Anterior, posterior, anddorsal to the pneumatic fossa, the centrum isfilled with pneumatic chambers. The smallsize, lack of regular branching pattern, andthinness of the septae dividing these chambersidentify them as camellae. The presence ofcamellae extends into the diapophyses andparapophyses. The camellate structure of thelatter is particularly highly developed. Theinternal structure of the cotyle and zygapoph-yses is unclear, possibly as an artifact ofscanning resolution, though it appears thezygapophyses are at least somewhat pneuma-tized. A large foramen above the diapophysisof the fourth vertebrae appears to communi-cate with the neural canal (see fig. 7d).

A few camellae are exposed on the surfaceon the dorsal side of the condylar neck of thefifth cervical (fig. 8). The bone at the neck ofthe condyle is very thin, as seen in the CTimagery. In several places, the outer layer ofbone has collapsed or been worn away,exposing camellae infilled with matrix. Asexposed on the surface, the camellae areirregularly shaped, giving the bone surround-ing them a honeycomb-like appearance. Theexposed septae are approximately 1 mm thick.Larger camallae with thin septae are alsoexposed on the lateral surface of the damagedsixth vertebral centra.

APPENDICULAR SKELETON

The sternal plate (fig. 9) exhibits a stronglyconcave lateral border, a feature seen inTitanosauria as well as in the rebbachisauridLimayosaurus tessonei (Calvo and Salgado,1995). The internal surface is concave poster-iorly and nearly flat anteriorly. The externalsurface is gently convex posteriorly and nearly


flat anteriorly. The sternal plate is greatlythickened at the anterolateral corner, but lacksthe anteroventral ridge developed in sometitanosaurs (Sanz et al., 1999).

The right tibia, fibula, and astragalus(fig. 10) are preserved in articulation withonly slight displacement. The hindlimb ele-ments bear a strong overall resemblance to

Fig. 5. Cervical 4 of IGM 100/1803 in (a) left lateral, (b) right lateral, (c) dorsal, (d) anterior, and (e)posterior view. For abbreviations, see text. Scale bar for (a), (b), and (c) 5 5 cm.


those of Gobititan shenzhouensis. The proximalface of the tibia is subcircular. The well-developed cnemial crest projects laterally,embracing the proximal end of the fibula.The greatest breadth at midshaft is in theanterolateral-posteromedial dimension, but atthe distal end the transverse breath is greatest.

The anterior and distal expansion of the tibiadoes not reach the degree seen in Opis-thocoelicaudia (Borsuk-Bialynicka, 1977).

The fibula extends distal to the tibia. Thefibula is broken near the distal end of thelateral trochanter, and the sigmoid appearanceof the shaft is slightly exaggerated by this

Fig. 6. Cervical 5 of IGM 100/1803 in (a) left lateral, (b) dorsal, and (c) right lateral view. This vertebrahas been deformed so that the centrum appears artificially dorsoventrally flattened and the right diapophysisis visible ventrally in view a. For abbreviations, see text. Scale bar 5 5 cm.


damage. The anterior and lateral trochantersarise approximately one-third of the bone’slength from the proximal end on the ante-rolateral surface of the shaft, as in Opis-thocoelocaudia. The trochanters are placed

more distally in Gobititan (You et al., 2003).The anterior trochanter is short, whereasthe lateral trochanter extends past the mid-point of the shaft and has a slight postero-distal slant. The anteromedial edge of the shaft

Fig. 7. Computed tomography images of cervical 4. (a) Lateral view of cervical 5 showing plane ofcoronal sections. Scale bar 5 5 cm. (b) Section through position 1. (c) Section through position 2. (d) Sectionthrough position 3. For abbreviations, see text.

Fig. 8. Dorsal view of the condyle of cervical 5. View (a) is unmodified; in view (b) the exposed camellaeare highlighted in dark gray.


develops into a sharp ridge near the distalend. The condyle is modestly expanded andhas a slightly convex distal surface.

The morphology of the astragalus is diffi-cult to observe because of its tight articulationwith the tibia. In anterior view it exhibits thetypical wedge shape of neosauropods (Wilsonand Sereno, 1998). It fails to cap the entiredistal end of the tibia, not reaching the medialborder. An ossified calcaneum is present. It isa small, globular bone with a highly rugosesurface texture and a flattened dorsal face.Although it remains debatable whether anossified calcaneum was present in theDiplodocidae (see Bonnan, 2000), this element

Fig. 9. Right sternal plate of IGM 100/1803 in(a) external and (b) visceral views. Scale bar 5 5 cm.

Fig. 10. Right tibia, fibula, astragalus, and calcaneum of IGM 100/1803 in (a) anterior and (b) posteriorviews. Scale bar 5 10 cm.


is known to be present in most sauropods,including Euhelopus and Gobititan. The lack ofa calcaneum in the articulated hindlimb of theholotype of Opisthocoelocaudia is consideredevidence of true absence (Borsuk-Bialynicka,1977) in that taxon.

PHYLOGENETIC ANALYSIS

A phylogenetic analysis was conductedutilizing a matrix of 234 characters and 29sauropod and outgroup taxa from an analysisby Wilson (2002) with the addition of Erketu.A single change was made to the originalmatrix, changing the coding for Camarasaurusfor character 90 from 0 to 1. This changereflects the observation of Tshuihiji (2004)that a median tubercle is present between theposterior cervical neural spines in Camara-saurus. A branch and bound search conductedin PAUP*4.0b10 yielded six most parsimoni-ous trees of 434 steps in which Erketuoccupied one of two positions: the sister taxonof Euhelopus or the sister taxon ofTitanosauria. The strict consensus of the sixmost parsimonious trees is presented infigure 11.

Erketu possesses two characters synapo-morphic of Titanosauriformes in this analysis:presence of presacral camellae and elongatecervical centra. Erketu possesses one characterof the more exclusive Somphospondyli: re-duced cervical neural arch lamination. In thesubset of trees supporting an Erketu +Titanosauria grouping, one character opti-mizes as an unambiguous synapomorphy ofthat clade: distally expanded tibia. A secondcharacter is ambiguously synapomorphic forthis clade: crescent-shaped sternal plates (un-known in Euhelopus). In the subset of treessupporting a Erketu + Euhelopus grouping,one unambiguous synapomorphy supportsthat clade: cervical centra higher than wide.

Because many important characters arescored unknown for Erketu, and also becausetaxon sampling of the Titanosauria is in-complete, the phylogenetic position presentedin figure 11 should be regarded as preliminary.As discussed below, we believe the associationof additional postcranial and dental materialwith Erketu may lead to a revised placementwithin the Titanosauria.

DISCUSSION

The size of the preserved hindlimb elementssuggests that Erketu, despite its great length,was of modest mass. The third, fourth, andfifth cervical vertebrae of Erketu are moreelongate relative to the tibia than are thecervical vertebrae of the long-necked sauro-pods Diplodocus carnegii, Omeisaurus tian-fuensis, Mamenchisaurus hochuanensis, andEuhelopus zdanskyi. Whether this indicatesa proportionally longer neck relative to trunkheight cannot be decided, as the forelimb,femur, and pes length and total cervicalvertebral count of Erketu are not known.

Various sauropod taxa achieve elongationof the neck by increasing the length of thecervical vertebrae, the number of cervicalvertebrae, or both. The long-necked titano-sauriform Brachiosaurus retains a plesio-morphic count of 13 cervical vertebrae, butexhibits great elongation of these elements,whereas Euhelopus increases its cervical verte-bral count to 17 without significant elongationof individual vertebrae. Although it is clearthat Erketu had elongate individual vertebrae,poor knowledge of the vertebral counts ofbasal titanosaurs precludes a phylogeneticallyinformed estimate of the total number ofcervical vertebrae.

The morphology of the cervical seriesreflects accommodation of stresses associatedwith extreme length. Osteological correlates ofthe extensor musculature are strongly devel-oped. The long extensor muscle M. ascendenscervicalis originates on the tuburculum ansaand overlaps one or more intermediate verte-brae before inserting on the torus dorsalis inliving birds and is hypothesized to have hadthe same attachment sites in sauropods (Wedeland Sanders, 2002). The hypertrophied pre-zygodiapophyseal lamina of Erketu is mostlikely associated with the attachment of thismuscle as it overlaps vertebrae (fig. 12). Thetorus dorsalis is likewise powerfully developedin Erketu. The tuberculum ansa is rugose andprojects posteriorly from the ansa costo-transversaria in cervical five, though its de-velopment is weaker in cervical four. Thedevelopment of features associated with M.cervicalis ascendens are weakest in the thirdcervical, which is not unexpected at the mostanterior portion of the neck.


Tshuihiji (2004) discussed bifurcation of theneural spine in sauropods and the extant ratiteRhea americana and concluded that the spacebetween the metapophyses housed extensor

ligaments, and possibly epaxial musculature aswell. Tshuihiji inferred that the rugose tuber-cles present posteriorly between the metapo-physes of Apatosaurus and Camarasaurus

Fig. 11. Strict consensus of six most parsimonious trees of 434 steps derived from analysis using thematrix, character orderings, and coding strategies of Wilson (2002) with the addition of Erketu. Bootstrapvalues were calculated in PAUP*4.0b10 from 1,000 replicates with TBR branch swapping. Bootstrap values.50 are shown above the node they refer to. Decay values calculated manually in PAUP*4.0b10 are shownbelow the node they refer to.

Fig. 12. Artificially articulated anterior cervical series of IGM 100/1803 illustrating the relative lengthincrease between vertebrae. The ansa costo-transversaria is reconstructed on the sixth vertebra. Note that allcervical ribs in this figure are incomplete; preserved ribs overlap at least one succeeding centrum, but havenot been reattached due to their fragile nature. The hypothesized position of M. ascendens cervicalis shownin white, with small arrows indicating attachment sites (after Wedel and Sanders, 2002: fig 2). Scale bar 550 cm.


served as insertion points for a ligamentsimilar to lig. elasticum interspinale in Rhea,which would serve to exert a supportingtensional force when the neck was tilted belowthe level of the ligament’s origin at theshoulder region. Tshuihiji noted that theleverage of this force would be increased bylowering the insertion point into the cleft ofa bifid neural spine, as opposed to an insertionon the dorsal surface of a single spine. Mediantubercles are not present in the knowncervicals of Erketu. However, the tuberclefirst arises on cervical ten in Apatosaurus(Gilmore, 1936), so this absence may be due tothe termination of this ligament system in themiddle part of the cervical column.

As elongation of the neck increases, thefunctional benefits of bifurcation should in-crease concomitantly. However, the distribu-tion of bifid neural spines in sauropods is notnecessarily correlated with neck elongation.Bifurcation is present in both long-necked(Diplodocus, Barosaurus) and shorter-necked(Dicraeosaurids, Apatosaurus) diplodocoidsand independently derived in the short-necked Camarasaurus, the long-neckedMamenchisaurus, and some titanosauriforms(Rapetosaurus, Euhelopus). Furthermore,many sauropods possessing extremely elon-gate necks, such as brachiosaurids andOmeisaurus, lack bifurcation.

Camellate internal structure evolved at leasttwice in sauropods, and is seen in the long-necked Mamenchisaurus and both long- andshort-necked titanosauriforms. The relativeweight reduction and structural strength ofcamellate vertebral design versus the camaratedesign seen in other sauropods is unclear,though initial analysis suggests equal bonereduction can be achieved with either of thetwo (Wedel, 2004).

The history of Asian sauropods is complex.The Jurassic sauropod fauna belong to a prim-itive, possibly endemic radiation, whereas theCretaceous fauna appears to be composedentirely of Titanosauriformes (Wilson, 2005).Whether Euhelopus belongs within theTitanosauriformes (Wilson, 2002) or outsideNeosauropoda (Upchurch, 1998; Upchurchet al., 2004) remains controversial. AlthoughTitanosauria is one of the most widespread andsuccessful clades of Cretaceous dinosaurs, the

interrelationships within this group are in-completely understood. Erketu occupies a posi-tion of potential importance in understandingthe transition from Titanosauriformes to thederived Titanosauria. Continuing work at BorGuve and other Gobi localities will helpelucidate the history of this group. Notably,cylindrical, unexpanded sauropod teeth areamong undescribed material recently collectedat Bor Guve. If these teeth can be confidentlyassigned to Erketu, it would make a strongargument that this taxon can be nested withinderived Titanosauria.

ACKNOWLEDGMENTS

We thank the 2002 AMNH-MAS fieldteam for collection of the specimen, JaneShumsky and James Klausen for skillfulpreparation, and Justin Sipla and JustinGeorgi for CT scans. We thank KristinaCurry Rogers for providing images ofRapetosaurus and Matt Wedel for discussions.A review by Jeff Wilson was most helpful inimproving this paper.

REFERENCES

Bonnan, M.F. 2000. The presence of a calcaneum ina diplodocid sauropod. Journal of VertebratePaleontology 20: 317–323.

Borsuk-Bialynicka, M. 1977. A new camarasauridsauropod Opisthocoelicaudia skarzynskii gen.n., sp. n. from the Upper Cretaceous ofMongolia. Palaeontologia Polonica 37: 5–64.

Calvo, J.O., and L. Sagaldo. 1995. Rebbachisaurustessonei sp. nov. A new sauropoda from theAlbian–Cenomanian of Argentina; new evi-dence on the origin of the Diplodocidae. Gaia11: 13–33.

Gilmore, C.W. 1936. Osteology of Apatosaurus,with special reference to specimens in theCarnegie Museum. Memoirs of the CarnegieMuseum 11: 175–300.

He., X., C. Li, and K. Cai. 1988. [The MiddleJurassic dinosaur fauna from Dashanpu,Zigong, Sichuan: sauropod dinosaurs. Vol. 4,Omeisaurus tianfuensis]. Chengdu: SichuanPublishing House of Science and Technology,143 pp.

Janesch, W. 1950. Die wirbelsaule von Bran-chiosaurus brancai. Palaeontographica (sup. 7)3: 27–93.

Kurzanov, S.M., and A.F. Bannikov. 1983. [A newsauropod from the Upper Cretaceous of


Mongolia]. Paleontologicheskii Zhournal 2:90–96.

Martin, V., E. Buffetaut, and V. Suteethorn. 1994.A new genus of sauropod dinosaur from theSao Khua Formation (Late Jurassic or EarlyCretaceous) of northeastern Thailand.Comptes Rendus de l’Academie des Sciences,Series II 319: 1085–1092.

McIntosh, J.S. 1990. Sauropoda. In D.B.Weishampel, P. Dodson and H. Osmolska(editors), The Dinosauria: 345–401. Berkeley:University of California Press.

Nowinski, A. 1971. Nemegtosaurus mongoliensisn.gen., n.sp. (Sauropoda) from the UpperCretaceous of Mongolia. PalaeontologicaPolonica 25: 57–81.

Osborn, H.F. 1924. Sauropoda and Theropoda ofthe Lower Cretaceous of Mongolia. AmericanMuseum Novitates 128: 1–7.

Powell, J.E. 1992. Osteologia de Saltasaurusloricatus (Sauropoda–Titanosauridae) delCretacico Superior del noroeste Argentino;In J.L. Sanz and A.D. Buscalioni (editors),Los Dinosaurios y su Entorno Biotico: Actasdel Segundo Curso de Paleontologia in Cuenca:165–230. Cuenca: Instituto ‘‘Juan de Valdes’’.

Powell, J.E. 2003. Revision of South Americantitanosaurid dinosaurs: palaeobiological, pa-laeobiogeographical and phylogenetic aspects.Records of the Queen Victoria Museum 111:1–173.

Russell, D.A., and Z. Zheng. 1993. A large mamen-chisaurid from the Jungar Basin, Xinjiang,People’s Republic of China. Canadian Journalof Earth Sciences 30: 2082–2095.

Sanz, J.L., J.E. Powell, J. le Loeuff, R. Martinez,and X.P. Suberbiola. 1999. Sauropod remainsfrom the Upper Cretaceous of Lano (NorthCentral Spain). Titanosaur phylogenetic rela-tionships. Estudios del Museo de CienciasNaturale de Alava 14: 235–255.

Sereno, P.C., A.L. Beck, D.B. Dutheil, H.C.E.Larsson, G.H. Lyon, B. Moussa, R.W. Sadleir,C.A. Sidor, D.J. Varricchio, G.P. Wilson, andJ.A. Wilson. 1999. Cretaceous sauropods fromthe Sahara and the uneven rate of skeletalevolution among dinosaurs. Science 286:1342–1347.

Shuvalov, V.F. 2000. The Cretaceous stratigraphyand paleobiogeography of Mongolia. In M.J.Benton, M.A. Shishkin, D.M. Unwin and E.N.Kurochkin (editors), The age of dinosaurs inRussia and Mongolia: 390–401. Cambridge:Cambridge University Press.

Tshuihiji, T. 2004. The ligament system in the neckof Rhea americana and its implication for the

bifurcated neural spines of sauropod dinosaurs.Journal of Vertebrate Paleontology 24(1):165–172.

Upchurch, P. 1998. The phylogenetic relationshipsof sauropod dinosaurs. Zoological Journal ofthe Linnean Society 124: 43–103.

Upchurch, P., P.M. Barrett, and P. Dodson. 2004.Sauropoda. In D.B. Weishampel, P. Dodsonand H. Osmolska (editors), The Dinosauria,2nd ed.: 259–322. Berkeley: University ofCalifornia Press.

Wedel, M.J. 2004. Skeletal pneumaticity in sauris-chian dinosaurs and its implications for massestimates. Journal of Vertebrate Paleontology24 (3) supplement: 127A.

Wedel, M.J., R.L. Cifelli, and R.K. Sanders. 2000.Osteology, paleobiology, and relationships ofthe sauropod dinosaur Sauroposeidon. ActaPalaeontologica Polonica 45(4): 343–388.

Wedel, M.J., and R.K. Sanders. 2002. Osteologicalcorrelates of cervical musculature in Aves andSauropoda (Dinosauria, Saurischia), with com-ments on the cervical ribs of Apatosaurus.Paleobios 22(3): 1–6.

Wilson, J.A. 1999. A nomenclature for the vertebrallamina of sauropod dinosaurs. Journal ofVertebrate Paleontology 19(3): 639–653.

Wilson, J.A. 2002. Sauropod dinosaur phy-logeny: critique and cladistic analysis.Zoological Journal of the Linnean Society 136:217–276.

Wilson, J.A. 2005. Redescription of the Mongoliansauropod Nemegtosaurus mongoliensisNowinski (Dinosauria: Saurischia) and com-ments on Late Cretaceous sauropod diversity.Journal of Systematic Palaeontology 3(3):283–318.

Wilson, J.A., and P.C. Sereno. 1998. Early evolu-tion and higher-level phylogeny of sauropoddinosaurs. Society of Vertebrate PaleontologyMemoir 5: 1–68.

Wiman, C. 1929. Die kreide-dinosaurier ausShantung. Paleontologia Sinica (C) 6: 1–67.

You, H., F. Tang, and Z. Luo. 2003. A new basaltitanosaur (Dinosauria: Sauropoda) from theEarly Cretaceous of China. Acta GeologicaSinica 77(4): 424–429.

Young, C.C. 1939. On a new Sauropoda, with noteson other fragmentary reptiles from Szechuan.Bulletin of the Geological Society of China 19:279–315.

Young, Z.J., and X.J. Zhao. 1972. Mamenchisaurushouchuanensis. Institute of Vertebrate Paleon-tology and Paleoanthropology Monographs,Series A (8): 1–32.


APPENDIX 1

ERKETU ELLISONI MEASUREMENTS (IN MILLIMETERS)

APPENDIX 2

CODINGS FOR ERKETU FOR CHARACTER MATRIX OF WILSON (2002) USED IN PHYLOGENETIC ANALYSIS

Vertebrae Centrum length Centrum height (at cotyle) Centrum width (at cotyle)

Axis 160 49 50

Cervical 3 268 60 46

Cervical 4 387 70 55

Cervical 5 489 57 (deformed) 85 (deformed)

Hindlimb Proximodistal length Distal breadth (anterior face)

Tibia 710 218

Fibula 750 —

. . . . . . . . . 10 . . . . . . . . . 20

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

. . . . . . . . . 30 . . . . . . . . . 40

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

. . . . . . . . . 50 . . . . . . . . . 60

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

. . . . . . . . . 70 . . . . . . . . . 80

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 1 1 0 ?

. . . . . . . . . 90 . . . . . . . . . 100

1 1 1 1 1 1 ? ? ? ? ? ? ? ? ? ? ? ? ? ?

. . . . . . . . . 110 . . . . . . . . . 120

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

. . . . . . . . . 130 . . . . . . . . . 140

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 1 0

. . . . . . . . . 150 . . . . . . . . . 160

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? 1 ? ?

. . . . . . . . . 170 . . . . . . . . . 180

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

. . . . . . . . . 190 . . . . . . . . . 200

? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?

. . . . . . . . . 210 . . . . . . . . . 220

? ? 1 1 1 ? ? 1 0 1 1 1 ? 0 ? ? ? ? ? ?

. . . . . . . . . 230 . . . .

? ? ? ? ? ? ? ? ? ? ? ? ? ?


Documents

Ksepca et al, 2006